Rethinking On-Policy Distillation of Large Language Models II: One Training Example Paper • 2609.04172 • Published 27 days ago • 103